ant-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Steve Loughran" <>
Subject Re: 'programming' extensions to ant discussion
Date Fri, 24 May 2002 07:40:16 GMT

----- Original Message -----
From: "charles_loboz" <>
To: <>
Sent: Thursday, May 23, 2002 10:45 PM
Subject: 'programming' extensions to ant discussion

>1. Things like <foreach> are popping up frequently in the list of
>requests and are (rightly imho) rejected by Ant designers. We have a
>conflict: the functionality is needed but we don't want Ant to became a
>programming language.

not a procedural one, at any rate; it is meant to be a declarative way of
describing how to build a project.

>2. It may possible to resolve this conflict. By 'programming' constructs
>most people mean 'imperative' programming - which conflicts with
>Ant/make _declarative_ style. The question is: how can we expand
>declarative style to provide the _functionality_ without introducing


>3. I do not propose any solutions, but a general direction. Ant
>frequently deals with _sets_ of objects - be it patterns, filesets,
>dirsets, properties... Let's generalize it a bit - introduce a _list_
>(set) as the basic 'type'. Each task can then assume that it's
>argument(s) is a list (possibly length one or zero) of objects. So, each
>task can then be used to process one or more objects. With some general
>build-in operations on lists (like filtering) we can obtain very
>declarative build patterns using very regular approach.

lists are the ideal way, something like [el1,elt2,elt3] makes for a good

>4. Admittedly, this is a shameful rip-off from Haskell or Scheme. This
>is not a proposal to implement Scheme in Ant :-) , but use concepts from
>more declarative world to regularize many Ant's tasks.

I was thinking more of prolog, myself

To unsubscribe, e-mail:   <>
For additional commands, e-mail: <>

View raw message